The role of stromal cells in epithelial–mesenchymal plasticity and its therapeutic potential

The most distinguishing characteristic of cells that experienced EMT is increased motility, which facilitates tumor cell invasion and metastasis. Indeed, cancers often lose epithelial markers (for example, E-cadherin, p53, and RB) and exhibit mesenchymal markers such as Vimentin, N-cadherin and Fascin1 (see Fig. 1) before becoming invasive or circulating tumor cells (CTCs), hence facilitating tumor spread. EMT promotes tumor metastasis by increasing cell motility and by increasing tumor cells’ stem-like state, therapeutic resistance, and immune evasion [31, 45, 46].

EMT is also strongly associated with tumor cell drug resistance. Some drug-resistant tumor cells have an EMT phenotype, and those undergoing EMT are resistant to chemoradiotherapy in breast cancer and other malignancies [47, 48]. Twist1 confers resistance to chemoradiotherapy by inhibiting apoptosis in hypopharyngeal cancer cell lines [49]. Snail1 enhanced the capacity of head and neck squamous cell carcinoma cell lines to repair DNA damage caused by cisplatin by up-regulating the excision and repair of cross-complementary HIF-1α protein [50]. Additionally, Twist or Snail conferred resistance to gemcitabine on pancreatic cancer cells by modulating the expression of drug-inactivated enzymes and/or transporters [51]. Due to the fact that EMT occurs in a variety of different cancer types through various pathways, cancer cells might develop resistance to chemoradiotherapy or drugs, and inhibiting EMT can improve treatment sensitivity. Taken together, these observations strongly imply EMT plays a paramount role in tumor progression and is, to some extent, responsible for the vast majority of drugs failures in the clinic.

Cancer stem cells (CSCs), also known as tumor-initiating cells (TICs) or CTC, are tiny subpopulations of tumor cells that play a pivotal role in carcinogenesis and progression, as these subpopulations that are undergoing EMT possess stem cell-like characteristics, namely be of the ability to self-renew and resistance to a range of therapeutic agents [31, 38, 52‐56]. It is noteworthy that the signaling pathways engaged during EMT are strikingly comparable to those that drive CSCs, including Wnt/β-catenin, TGF-β/Smads, and other signaling pathways that are required for CSC self-renewal and maintenance [38]. Accordingly, these findings explain why a minute proportion of CSCs are produced during EMT.

Epithelial cells are normally polarized apical-basal axis and are connected by tight junctions, adherens junctions, desmosomes, and hemidesmosomes. E-cadherin, a core molecule on the cell surface, is responsible for adhesion and controlling and/or maintaining cells’ epithelial identity or phenotype [38, 57]. This link is critical for epithelial tissue’s structural integrity and solid tumor’s formation. Following EMT activation, the characteristic polygonal and pebble-like shape of epithelial cells or cobble-stone cell islands phenotype is lost, resulting in cells being more readily separated from epithelial tissues. A recent study corroborated that the cellular adhesion and polarity of colorectal carcinoma were diminished when EMT occurs. Accordingly, a drop in the expression of epithelial markers was observed [58]. On the other side, interstitial features are acquired when cells develop a fusiform mesenchymal shape, which is accompanied by a rise in the mesenchymal markers N-cadherin, Fascin1, Fibroblast specific protein 1 (Fsp1) and Vimentin [42, 59‐63] (Fig. 1).

Mesenchymal cells exhibit mobility and migration due to interstitial markers and changes in cell shape. Due to the gain of front-rear polarization nature instead of apical-basal polarization and the absence of intercellular connections, individual mesenchymal cells may migrate across the extracellular matrix (ECM). Reduced E-cadherin expression and increased N-cadherin expression decrease cells’ adhesion ability and increase cell mobility, allowing cells to migrate and invade, which are indicative of EMT [64, 65]. It is not difficult to observe that EMT fundamentally transforms cell phenotypes by altering the expression of epithelial and mesenchymal phenotypic markers.

On the other hand, mesenchymal cells may return to an epithelial condition through the reversal of EMT process, namely the MET. These changes are intimately connected with tumor differentiation [66]. By reducing E-cadherin expression, initial cancer cells lose their cellular adhesion and gain mesenchymal features, converting them into migratory and invasive cancer cells. These cancer cells will pierce the basal membrane, enter the blood through blood arteries, and then overflow into the area of metastasis, creating micrometastasis on the target organ’s parenchyma [67‐69]. Once reaching the metastatic location, tumor cells undergo MET to revert to their epithelial origin, forming secondary malignant tumors. The back-and-forth conversion between EMT and MET will eventually kill the cancer patients who are bearing a tumor mass of approximately 1 kg or harboring 10¹² to 10¹³ cancer cells [70]. This diagram depicts the whole process of tumor cell migration through EMT (Fig. 2).

When cancer cells and tumor stem cell-like cells undergo EMT, they display immunosuppressive and immune-resistant properties. Cancer cells may interact with various immune cells and other stromal cells in tumor tissue, resulting in the formation of an immunosuppressive microenvironment. Indeed, snail-induced EMT may promote melanoma progression by altering T cell-mediated immune suppression. Additionally, EMT-producing cells may impede the killing function of cytotoxic T cells by disrupting immunological synapses. EMT may potentially help in immune evasion by boosting the expression of immunological checkpoint proteins on tumor cells, such as programmed-cell death ligand 1/2 (PD-L1/2) and B7-H3 [71‐74].

Among the cytokines generated by CAFs, the most extensively researched is TGF-β, which plays a critical role in developing many types of malignancies [97, 98]. Mesenchymal fibroblasts may release TGF-β and stromal cell-derived factor-1 (SDF1), which function in concert through distinct signaling pathways to sustain mesenchymal fibroblast differentiation into myofibroblasts during cancer development. TGF-β levels were much greater in myofibroblasts than in normal stromal fibroblasts [99].

TGF-β, which CAFs release, influences EMT by modulating gene transcription in the nucleus through many distinct signaling pathways. The TGF-β/Smads signaling pathway is critical for regulating EMT-related genes in cancer cells among several signaling pathways. TGF-β is recognized by two distinct receptor types: type I and type II receptors (TRI and TRII, respectively). TGF-β dimers bind to its receptors (TRI and TRII) on the cell surface, inducing TRII phosphorylation followed by phosphorylating TRI through its serine and threonine kinases domain. Subsequently, the receptor-phosphorylated Smad2/3 (R-Smads) is recruited and phosphorylated by active type I receptors, and these receptor-activated Smads (R-Smads) form protein complexes with the ubiquitous Smad4 (Co-Smad). The activated Smads complex is transported to the nucleus and cooperates with other transcription factors to co-regulate the transcription of target genes [100‐102] (Fig. 3). Additionally, TGF-β may exert its effect via regulating MAPK, PI3K/Akt, and other cancer-related signaling pathways through non-classical signaling pathways [103]. Similarly, TGF-β is plentiful in CAFs conditioned media and may promote EMT in bladder cancer cells by activating Smad2, which activates classical TGF-β signaling [104, 105]. TGF-induced EMT resulted in the overexpression of genes associated with Snail1, ZEB1, and ZEB2. It also leads to the down-regulate of the epithelial marker E-cadherin and up-regulation of the mesenchymal markers such as vimentin and N-cadherin, ultimately EMT transformation [105].

IL-6, an inflammatory factor, is another key cytokine released by CAFs that promotes EMT in tumor cells. CAFs induce EMT by secreting IL-6 and promoting cancer cell migration [96]. Signals travel from outside the cells to the nucleus following JAK2/STAT3 signaling pathway, which is activated when IL-6 binds to the receptor on the cell membrane. Specifically, the IL-6 signal by activating JAK2 protein kinase mediates the phosphorylation of STAT3, which is then translocated from the cytoplasm into the nucleus to regulate the expression of EMT-related and other genes [96, 106]. Similarly, IL-6 may alter the phenotype of human peritoneal mesothelial cells (HPMCs) in vitro from pebble-like to fibroblast-like, which was an EMT feature [107]. Also, IL-6 produced from CAFs may develop resistance to cisplatin in non-small cell lung cancer (NSCLC) through promoting EMT [108]. According to a study, CAFs boost TGF-β mediated EMT in ovarian cancer by increasing IL-6 production through the JAK2/STAT3 pathway, which suppresses apoptosis and results in higher resistance to paclitaxel [109]. It is clear from these findings that the mechanism by which CAFs secrete IL-6 to induce EMT in tumor development might enhance tumor incidence and spread as well as tumor medication resistance. More CAF-secreted IL-6 is expected to enhance cancer cell migration and EMT-specific activities shortly.

Beyond the factors that were mentioned above, other growth factors and chemokines secreted by CAFs can also induce EMT: in pancreatic and breast cancer, VEGF produced by CAFs causes EMT by increasing the expression of EMT-related genes such as Snail and Twist [110, 111]; matrix metalloproteinases (MMPs) such as MMP2, MMP3 and MMP9 generated by CAFs degrade E-cadherin through the Rac1b/ROS pathway to fuel β-catenin entry into the nucleus to induce EMT and improve HCC cell motility [111, 112]; gastric cancer cell metastasis was also increased through the C-MET (HGF receptor) pathway activation when hepatocyte growth factor (HGF), secreted by CAFs, was binding to C-MET [113]. Indeed, the HGF ligand triggers EMT in tumor cells by activating many intracellular signaling pathways [114].

In addition to these mentioned growth factors and chemokines, other secretions also generated by CAF, including Sonic Hedgehog (Shh) [115], Netrin-1 [116], Jag/Dll [117], Wnt [118], 17 β-estradiol (E2) [119] and α-smooth muscle actin (α-SMA) [120], can also favour EMT activation in tumor cells.

Induction by TAMs is another dramatic mechanism for EMT. As a highly malleable immune cell, macrophages are critical for immunological metabolism, chronic inflammation, and tissue homeostasis [129, 130]. TAMs are macrophages that move to tumor stroma and are often the most numerous immune cells highjacked by cancer cells to enter into the TMEs to distort their properties from anti- to pro-cancer [131] (Fig. 4). They play a role in the genesis of cancer, the generation of drug resistance, and immunological escape in a variety of malignant tumors, and are linked with a poor prognosis [130, 132]. With respect to the pro-cancer role of TAM in TMEs, a report shows TAMs accelerate cancer progression by increasing stromal remodeling, suppressing adaptive immunity, and boosting angiogenesis and EMT [133].

TAMs have distinct behaviors in various malignancies, often manifesting as the polarization of anti-inflammatory and pro-tumor phenotypes, including the M1 and M2 subtypes [134‐136]. M1-type macrophages have been shown to slow down tumor growth, but M2 macrophages release a range of cytokines that induce EMT and immunosuppression, TGF-β, IL-1β/6/10, CXCL10, HIF-1, and VEGF are a few of these cytokines [137‐140]. These cytokines act as a suppressor of adaptive immunity during the onset of EMT. M2 macrophages and regulatory T cells (T-regs) are more abundant in tumor stroma formed from mesenchymal cell lines than epithelial cell line-derived cancers [71]. Under hypoxia, Hif-1α stimulates the release of CCL-20 by mesenchymal cancer cells, which in turn increases the expression of indoleamine 2,3-dioxygenase (IDO) in TAMs, ultimately impairing the function of CD4⁺ and CD8⁺ T cells in hepatocellular carcinoma [141]. Likewise, inhibition of mTORC1/mTORC2 may suppress EMT and down-regulate TAM recruitment and PD-L1 expression, therefore boosting immunity against lung cancer [142]. The M2 subtype of macrophages was also encouraged to polarize in gastric cancer-derived mesenchymal stromal cells (GC-MSC). The GC-MSC-induced M2-type macrophages then increased the expression of interstitial marker genes vimentin, fibronectin, and Slug. Conversely, E-cadherin expression was dramatically reduced. Hence, GC-MSC-induced M2 macrophages promote gastric cancer cell EMT and thus tumor spreading. Ontogeny-induced M2-type macrophages play a critical role in TAM formation and immunosuppression [143]. Thus, inhibiting anti-tumor M1-type macrophages skew to pro-tumor M2-like macrophages in TME may suppress tumor progression by aiming to EMT and provide a novel target for cancer treatment.

Several different mechanisms exist for TAMs to produce cytokines to drive tumor EMT. In a paracrine way, similar to CAFs, TAMs release Wnt2b in hepatocellular and thyroid carcinoma cells, inducing EMT [144, 145]. TAM may also release tumor necrosis factor (TNF), and the synergistic effect of TNF and TGF-β has been shown to enhance EMT in colon cancer [146].

Mesenchymal-like breast cancer cells attract TAMs by the secretion of granulocyte–macrophage colony-stimulating factor (GM-CSF), while TAMs recruited by surrounding tumor cells produce CCL18. Therefore, metastasis is facilitated by inducing EMT, and patient survival is, no doubt, shortened [147]. Additionally, IL-6 produced by TAMs operates on the COX-2/PGE2 signaling pathway and promotes EMT by stimulating the transfer of β-catenin from the cytoplasm to the nucleus to regulate the EMT-related gene expression [148].

Bone marrow-derived suppressor cells (MDSCs) are a member of TIICs and cause EMT activation. Although MDSCs are suppressor cells, De Cicco et al. found that MDSCs play an anti-cancer role through the excretion of H₂S, a gasotransmitter, in the early-stage [149], similar to TAM (see Fig. 4). MDSCs are immature bone marrow cell types that have been shown to induce EMT and accelerate non-small cell lung cancer metastasis via the CCL11-ERK/AKT-EMT axis [150]. A recent work showed that MDSCs are attracted into primary tumors by myeloid-specific chemical attractant SPARC, an extracellular matrix (ECM) component, thereby inducing the EMT programmes in high-grade breast cancer cells [151]. Once recruited, MDSCs cluster around the original tumor, promoting EMT, creating cancer stem cell-like cells, and disseminating cancer cells through the activation of TGF-β, COX2, and the secretion of EGF and HGF [152].

MDSCs, as variable myeloid cell morphologies, may contribute to immunosuppression in TME through EMT. MDSCs suppress a variety of T cells, nNK cells, and dendritic cell activities [153]. MDSCs have osmotic and inhibitory properties regulated by the inducible enzymes CXCL2 and COX-2 in mammary and colorectal cancer [151, 154], respectively. Snail increased CXCL2 expression through NF-kB signaling and encouraged MDSC recruitment into the tumor microenvironment, suppressing CD8⁺ T cells and encouraging ovarian tumor development [155]. Similarly, upregulation of β-catenin/TCF4 and COX-2 in nasopharyngeal cancer increases MDSC-mediated EMT. Additionally, COX-2 has been shown to regulate the interaction of cancer cells with MDSCs, hence promoting metastasis [156]. As such, MDSC-induced EMT contributes to tumor cell migration and this is comparable to how CAFs-induced EMT plays a role in tumor evolution.

Immune cells also contribute to EMT and tumor metastasis. Numerous immune cells accumulating in the tumor stroma interact with neighboring cancer cells, reactivating latent EMT processes. Although the general view is that the cytotoxic CD8⁺ T cells are of anti-tumor capabilities and serve as the mainstay of immunotherapy, this subset of cells can also induce EMT and increase tumor metastasis as representative immune cells in immune surveillance and immune education. For example, the melanoma-associated antigen C3-facilitated EMT and metastasis in esophageal squamous cell carcinoma are mediated by PD1⁺ CD8⁺ T cells infiltrating the tumor [157]. In addition, CD8⁺ T cells were discovered to promote EMT in vivo, resulting in the formation of mesenchymal breast cancer cells with stem-like characteristics [158] (Fig. 4).

CD4⁺ T cells, in addition to CD8⁺ T cells, may initiate the EMT process. Importantly, reduced interferon-gamma expression was seen in PanINs that were densely surrounded by CD4⁺ T cells, in contrast to PanINs that were not tightly surrounded by CD4⁺ T cells [159]. Furthermore, co-culture with activated effector CD4⁺ T cells in vitro results in the down-regulation of E-cadherin in pancreatic ductal epithelial cells. The epithelial cells acquire a fusiform mesenchymal appearance with vimentin and ZEB1 expression [160]. Additionally, researchers discovered that CD4⁺ T cells, the primary source of IL-6, triggered the EMT process in clear-cell renal cancer [161]. Although the mechanism by which T cells induce EMT remains unknown, cytokines generated by activated effector T cells (e.g., IL-6, TNF, and TGF-β) are known to promote EMT.

As forementioned, NK cells, neutrophils, and dendritic cells are part of TIICs, which, together with suppressive factor of TME such as TGF-β constitute the TIME that plays a predominant role in tumorigenesis and progression. An accumulating body of evidence disclosed that TIICs play both pro- and anti-cancer roles during tumor development, akin to their partner stromal cells. The pro- or anti-cancer activity of TIICs also depends on the stage of tumor, as shown in Fig. 4. Considering the auxiliary role of dendritic cells in immune activities, we will, therefore, briefly discuss the functions of NK cells and neutrophils in tumor progression here.

As an important player in the TME, NK cell undoubtedly plays a key role in tumor evolution (Fig. 4). Yu’s group demonstrated that m6A reader protein YTHDF5-mediated anti-cancer potential of NK cells primarily depends on IL-5 stimulation to sustain NK cell survival and expansion [162]. Moreover, a recent work by Peng et al. showed that down-regulation of NK cell-derived IFNγ and TNF was positively associated with short survival of patients with gastric cancer [163]. Furthermore, a study in hepatocellular carcinoma also suggested that significantly decreased production of IFNγ due to fewer NK cell within tumor was responsible for poor survival [164]. Finally, we also found the anti-cancer effect of NK cell was mediated by the secretion of IFNβ [165]. Interestingly, research has also found NK cells can reduce their IFNγ production via up-regulation of CTLA4, which is accompanied by tumor development, thereby promoting tumor progression [166]. Similarly, Thacker et al. revealed breast cancer stem cells were activated by NK cells; however, abrogation NK or blocking NK-secreted Wnt prevented breast cancer progression [167]. These results indicate that NK cell plays both pro- and anti-cancer in tumor evolution, and the tumor stage controls specific tumor-promoting or tumor-suppressing function.

It has long been known neutrophils are the most abundant cells in the blood, meaning neutrophils may play conflicting roles in disease, particularly in cancer. As anticipated, a large body of research has unmasked mechanisms by which neutrophils fight cancer cells through secretions. For instance, IFNβ and IFNγ are two significant factors in TME that have been found to enhance the cytotoxicity of neutrophils, enabling them to combat early-stage breast cancer, lung cancer, and melanoma in both mouse and clinical models [168‐170]. In addition to IFNβ and IFNγ, an investigation of early-stage lung cancer samples also found that GM-CSF is another enhancement factor that promotes neutrophils’ anti-cancer capacity [170]. Furthermore, a growing number of evidence supports gasotransmitter NO can cause cancer cell apoptosis. Finisguerra et al., using a mouse modle, found NO produced by neutrophils can directly induce cancer cell death, and this cytotoxicity depends upon the expression of the HGF receptor, suggesting HGF receptor may be the ‘Achilles’ heel’ of cancer [171]. Additionally, Lev Becker’s group recently studied the role of neutrophils in several types of cancer cells, including ovarian, breast, lung, prostate cancer and melanoma cells and found that neutrophil-released specific protease ELANE induces cancer cell apoptosis via elevated levels of ROS and activation of CD8⁺ T-cells, suggesting neutrophils may possess a broad anti-cancer function [172]. However, clinical data suggests a high number of neutrophils in patients with advanced cancer is often associated with poor survival, indicating neutrophils may act as an accessory to the mastermind to assist tumor development. A great deal of study has been conducted to support this and answer how neutrophils corrupt to conspire to crime. For example, the roles of HGF and ROS, secreted from neutrophils, were examined, and their promotion role in tumor evolution was verified [171, 173], further suggesting neutrophil plasticity could be co-opted as a breakthrough in tumor therapy. This may be due to the higher concentration of PGE2 and TGFβ in late-stage, as well as other players in TME that have been demonstrated to accelerate tumor aggressiveness [174, 175]. Importantly, the production of PGE2 and TGFβ can both stem from cancer cells and non-canner cells, creating a reciprocal vicious cycle that promotes tumor progression. Unlike Lev Becker’s report, Maas et al. attempted to figure out whether neutrophils likely play the same role in brain tumor as they do in non-brain tumor, acting both pro- and anti-cancer via ROS. They found that ROS derived from neutrophils was decreased by glyoxalase 1 and IL19 to prolong neutrophile survival to prevent glioma cell from death [176]. The rational explanation is that the abundance and diversity of the components in the brain TME are less than their non-brain TME counterpart. Another good example of neutrophil pro-tumor is via inflammatory factor IL17a. More recently, Khalid et al. unveiled that increased IL17a relates to short survival. Further study manifested that the pro-tumor function of IL17a was exerted by recruiting neutrophils [177], strongly reflecting neutrophils played a key role in promoting cancer. Together, these data suggest neutrophils play a contradictory role during tumor evolution, as illustrated in Fig. 4. Exploring how to convert the pro-tumorigenic role to a tumor-suppressive one is a hotspot field in future investigation.